Closed Bug 386067 Opened 17 years ago Closed 16 years ago

session restore nag on closing random window (ok, yeah, the last one) even when window close confirmations disabled

Categories

(Firefox :: General, defect)

defect
Not set
normal

Tracking

()

RESOLVED DUPLICATE of bug 386628

People

(Reporter: tuukka.tolvanen, Unassigned)

References

(Depends on 1 open bug)

Details

firefox trunk 2007-06-27-11Z linux

str:
 1) open several browser windows with several tabs each
 2) close each window via window close button

expected:
 a) confirmation dialog on every window close if browser.tabs.warnOnClose true
 b) no confirmation dialog on any window close if browser.tabs.warnOnClose false

result: (a) works, (b) doesn't. I close a window when I'm done with the documents / session histories in it; an arbitrary one of the windows is now being somehow magical and gives me a dialog to save its documents for when I invoke another browser window. That's just in no way relevant to what I'm trying to do though, which is to destroy that window like I just did a bunch of others without any nags. I'll often (usually, even always?) want to save session on File -> Quit however...

I suppose the quit/session configmation dialog just shouldn't be displayed unless browser.tabs.warnOnClose is enabled; the folks, for whom it makes sense to ask for session save when closing a single window, wouldn't tend to have any reason to set that false, right?
There are two prefs now: browser.tabs.warnOnClose and browser.tabs.warnOnQuit. Having .warnOnClose set while not .warnOnQuit makes sense as you can have the last closed window restored while all others not (in that way, the last window _is_ special).

Having .warnOnQuit without .warnOnClose may make sense for people who don't care about losing data and just may occasionally want to restore one. The UX of the Quit prompt should be improved anyway on some of the other dependencies of bug 383760, so this issue might be vane in a few weeks.
Component: Session Restore → General
Depends on: 383760
OS: Linux → All
QA Contact: session.restore → general
Hardware: PC → All
Let me throw a table at you to pinpoint the issue in that context:

warnOn     action
---------- ----------------------
close quit close win/last  quit
----- ---- -------------- -------
on    on   window/session session
on    -    window/window  window
-     on        -/session session
-     -         -/-       -

See the -/session? That's a useless confirmation because the user can't trust that it's going to be there unless they manually enumerate their windows before taking the close action. So depending on what you want, in that case it's either an invitation to pick between extra work or dataloss if you try to trust it to allow restoring that window, or a useless distraction if you just want to close a window as any other. The other scenarios look fine I think.
Summary: session store nag on closing random window (ok, yeah, the last one) even when window close confirmations disabled → session restore nag on closing random window (ok, yeah, the last one) even when window close confirmations disabled
sorry, warnOnClose && !warnOnQuit && quit -> no confirmations, but luckily that's beside the point :)
Shouldn't the second row above read:
on    -    window/-       -

Otherwise we could do with .warnOnClose alone and use the following matrix:

warnOn  action
------- ----------------------
close   close win/last quit
------- -------------- -------
on      window/session session
-       -/-            -
> Shouldn't the second row above read:
> on    -    window/-       -

No, surely if you want to be warned about closing multiple tabs in one window close, it should apply to the last window just as any other (and it does).

Regarding not having a separate warnOnQuit, that wouldn't support not having nags on any window close but having the explicit-quit session nag -- which is probably what I want, so that I can easily restore the whole session or nuke the whole session on explicit quit. I'm not sure if that's important, I can just set it to restore always and try to cope with the browser loading content slower and throwing a bunch of random old documents at me if I happen to have closed the last window during some usage...

...but since the two prefs do exist now I'd rather just lose the session nag on !warnOnClose && warnOnQuit && close, and keep the session nag on !warnOnClose && warnOnQuit && quit.
Looking at several bugs (bug #433123, bug #422040, bug #412640, bug #419009), it seems that not everybody is comfortable with the semantics of browser.warnOnQuit and browser.tabs.warnOnClose. I wonder if decoupling them a little more and giving them the semantics I describe in bug #384907 comment #21 would be a step in the right direction. Any thoughts?
IMO warnOnQuit and warnOnClose shouldn't interfere. warnOnQuit warns you if you want to be asked every time to save session on exit. warnOnClose is to avoid to close accidentally a Fx window with many tabs.

See my proposal at Bug 443396 to make this more clear and simple to set.
Keywords: qawanted
Whiteboard: [invalid? see comment 7]
This was FIXED in bug 419009. Duping to the other such bug (which I happened to find first), though.
Status: NEW → RESOLVED
Closed: 16 years ago
Keywords: qawanted
Resolution: --- → DUPLICATE
Whiteboard: [invalid? see comment 7]
You need to log in before you can comment on or make changes to this bug.